Novel Method of Mining Classification Information for SVM Training

作     者：SHEN Fengshan ZHANG Junying YUAN Xiguo SHEN Fengshan,ZHANG Junying,YUAN Xiguo School of Computer Science and Engineering,Xidian University,Xi’an 710071,Shaanxi,China

作者机构：School of Computer Science and Engineering Xidian UniversityXi＇an 710071 Shaanxi China 

出 版 物：《Wuhan University Journal of Natural Sciences》 (武汉大学学报（自然科学英文版）)

年 卷 期：2011年第16卷第6期

页      面：475-480页

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 081201[工学-计算机系统结构] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Supported by the National Natural Science Foundation of China (61070137,60933009) the Science and Technology Research Development Program in Shaanxi Province of China (2009K01-56) 

主　　题：support vector machine (SVM) classification information incremental training candidate support vector 

摘      要：Support vector machine （SVM） is an important classi- fication tool in the pattern recognition and machine learning community, but its training is a time-consuming process. To deal with this problem, we propose a novel method to mine the useful information about classification hidden in the training sample for improving the training algorithm, and every training point is as- signed to a value that represents the classification information, respectively, where training points with the higher values are cho- sen as candidate support vectors for SVM training. The classifica- tion information value for a training point is computed based on the classification accuracy of an appropriate hyperplane for the training sample, where the hyperplane goes through the mapped target of the training point in feature space defined by a kernel fimction. Experimental results on various benchmark datasets show the effectiveness of our algorithm.